Explore the vast range of titles available.

Human Activity Recognition (HAR) has been studied extensively, yet current approaches are not capable of generalizing across different domains (i.e., subjects, devices, or datasets) with acceptable performance. This lack of generalization hinders the applicability of these models in real-world environments. As deep neural networks are becoming increasingly popular in recent work, there is a need for an explicit comparison between handcrafted and deep representations in Out-of-Distribution (OOD) settings. This paper compares both approaches in multiple domains using homogenized public datasets. First, we compare several metrics to validate three different OOD settings. In our main experiments, we then verify that even though deep learning initially outperforms models with handcrafted features, the situation is reversed as the distance from the training distribution increases. These findings support the hypothesis that handcrafted features may generalize better across specific domains.

With neutron diffraction, the local stress and texture of metallic components can be analyzed non-destructively. For both, highly accurate positioning of the sample is essential, requiring the measurement at the same sample location from different directions. Current sample-positioning systems in neutron diffraction instruments combine XYZ tables and Eulerian cradles to enable the accurate six-degree-of-freedom (6DoF) handling of samples. However, these systems are not flexible enough. The choice of the rotation center and their range of motion are limited. Industrial six-axis robots have the necessary flexibility, but they lack the required absolute accuracy. This paper proposes a visual servoing system consisting of an industrial six-axis robot enhanced with a high-precision multi-camera tracking system. Its goal is to achieve an absolute positioning accuracy of better than 50μm. A digital twin integrates various data sources from the instrument and the sample in order to enable a fully automatic measurement procedure. This system is also highly relevant for other kinds of processes that require the accurate and flexible handling of objects and tools, e.g., robotic surgery or industrial printing on 3D surfaces.

Neutron diffraction is a powerful non-destructive volumetric evaluation method for the analysis of the internal stress state in components processed by laser powder bed fusion (LPBF). High cooling rates and heterogeneous distribution of temperature during additive manufacturing lead to large residual stress fields. Residual stresses developed during the building process have unquestionably an important influence on the mechanical performance and potentially lead to delamination from the support structures, shape distortion but also crack formation. In the present work, neutron measurements have been carried out on cube-shaped samples prepared by LPBF from a Ti-6Al-4V powder bed. A series of miscellaneous positions (center, edge, and corner) over three different depths (close substrate, middle, and close surface) have been analyzed by neutron diffraction so as to systematically characterize the full stress tensor. The influence of shear stresses and second-order residual stresses on the stress tensor analysis is also discussed in this work.

The study of Domain Generalization (DG) has gained considerable momentum in the Machine Learning (ML) field. Human Activity Recognition (HAR) inherently encompasses diverse domains (e.g., users, devices, or datasets), rendering it an ideal testbed for exploring Domain Generalization. Building upon recent work, this paper investigates the application of regularization methods to bridge the generalization gap between traditional models based on handcrafted features and deep neural networks. We apply various regularizers, including sparse training, Mixup, Distributionally Robust Optimization (DRO), and Sharpness-Aware Minimization (SAM), to deep learning models and assess their performance in Out-of-Distribution (OOD) settings across multiple domains using homogenized public datasets. Our results show that Mixup and SAM are the best-performing regularizers. However, they are unable to match the performance of models based on handcrafted features. This suggests that while regularization techniques can improve OOD robustness to some extent, handcrafted features remain superior for domain generalization in HAR tasks.

The effect of the heat treatment on the residual stresses of welded cladded steel samples is analyzed in this study. The residual stresses across the plate’s square sections were determined using complementary methods; applying diffraction with neutron radiation and mechanically using the contour method. The analysis of the large coarse grain austenitic cladded layers, at the feasibility limits of diffraction methods, was only made possible by applying both methods. The samples are composed of steel plates, coated on one of the faces with stainless steel filler metals, this coating process, usually known as cladding, was carried out by submerged arc welding. After cladding, the samples were submitted to two different heat treatments with dissimilar parameters: one at a temperature of 620 °C maintained for 1 h and, the second at 540 °C, for ten hours. There was some difference in residual stresses measured by the two techniques along the surface of the coating in the as-welded state, although they are similar at the welding interface and in the heat-affected zone. The results also show that there is a residual stress relaxation for both heat-treated samples. The heat treatment carried out at a higher temperature showed sometimes more than 50% reduction in the initial residual stress values and has the advantage of being less time consuming, giving it an industrial advantage and making it more viable economically.

Machining part distortion is due to residual stresses induced by previous manufacturing processes. This study aims to evaluate the influence of machining conditions on AISI 316L plate distortion. Therefore, a special experimental device with force sensors integrated in the clamping system and numerical model of distortion were developed. Residual stresses due to previous machining processes were measured using a layer removal method and neutron diffraction technique. Then, distributions of these residual stresses were integrated in a developed model of machining distortion, which considers the clamping and machining sequence effects after each stage of the toolpath. A comparison of the experimental and numerical results revealed that the finite element method can adequately predict machining distortion. The results also suggest that clamping and machining sequence can affect part distortion.

The Commission of the Black Book on the Fascist Regime was created in 1977, in Portugal, to investigate the abuses perpetrated during the dictatorship that started in May 1926 and ended with the military coup of April 25, 1974. It was proposed by Mário Soares, the former socialist Prime Minister, as a way to fight the resurgence of fascist ideologies. The National Commission on the Disappeared (CONADEP), established in Argentina in 1983, is known as the first truth commission and comparative studies do not recognize the existence of a commission in Portugal. This investigation states, however, that the Commission of the Black Book must be considered a pioneering truth commission in the transitional justice processes of the third wave of democratizations. A Comissão do Livro Negro Sobre o Regime Fascista foi criada em Portugal, em 1977, para investigar os abusos cometidos durante o regime ditatorial inaugurado em maio de 1926 e derrubado pelo golpe militar de 25 de abril de 1974. Surgiu por proposta do então Primeiro-Ministro socialista Mário Soares como forma de combate ao ressurgimento de ideologias fascistas. A Comissão Nacional sobre Pessoas Desaparecidas (CONADEP), instaurada na Argentina em 1983, é tida como a precursora entre as comissões de verdade e os estudos comparados não reconhecem a existência de uma comissão em Portugal. Esta análise demonstra, contudo, que a Comissão do Livro Negro deve ser considerada uma comissão de verdade pioneira nos processos de justiça de transição da terceira vaga de democratizações.

Since the emergence of hypnosis, we have witnessed a multiplication of psychotherapies, which have different backgrounds and aims. The omnipresence of psychotherapy leads us to an inevitable question: what is psychotherapy? In this article, we analyse the concept of mental disorder and how psychotherapy works, underlining three mechanisms: influence, polyphonic dialogue and play. Focusing on the therapeutic dialogue, we explore what is told during therapy and how, what is done while telling, and how dialogue can create new narratives and new meanings, highlighting the importance of influencing the patient on changing the symptomatic behaviour. We also consider how the multiple voices of the patient, therapist and others can generate an alternative to the monologic discourse of the disease. While the psychiatric illness may indicate a sterile dialogue which often expands the pathology, communicating in a different and active way can create new and healthier meanings. Therefore, one of the therapy’s aims is to influence the patient, throughout a dialogic and playful conversation, to gain freedom from disease.

The Institute of Space Systems is currently developing a deorbit module based on thermal arcjet technology to allow fast orbit decay at end-of-life, with a focus on megaconstellation satellites. By employing additive manufacturing with tungsten, improved nozzle geometries can lead to a gain in overall performance. However, reproducibility is an ongoing concern for additively manufactured parts. Together with the Heinz Maier-Leibnitz Zentrum (MLZ) and the Budapest Neutron Centre of the Centre for Energy Research, a study was conducted scanning additively manufactured arcjet nozzles prior to and after standardized operation via neutron computed tomography. The results show a drop in performance over time, which can be related to changes in the constrictor geometry. Furthermore, cavities created during manufacturing can significantly influence operation.

Structural and spatial aspects of cell degradation are studied using a combination of diffraction-and imaging-based tools applying laboratory X-rays, neutron scattering and synchrotron radiation with electrochemical and thermal characterization. Experimental characterization is carried out on cylindrical cells of 21700-type, where four regimes of cell degradation are identified, which are supplemented by an increased cell resistance and surface temperature during cell operation. The amount of intercalated lithium in the fully charged anodes in the fresh and aged states is determined by ex situ X-ray diffraction radiography and in situ X-ray diffraction computed tomography. The qualitatively similar character of the results revealed a loss of active lithium along with the development of a complex heterogeneous distribution over the electrode stripe.